Abstract: Transformers are widely used in computer vision areas and have achieved remarkable success. Most state-of-the-art approaches split images into regular grids and represent each grid region ...
Abstract: Accurate acquisition of 3-D human joint poses holds significant implications for tasks such as human action recognition. Monocular single-frame 2-D -to-3-D pose estimation focuses on ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results